LE JOURNAL CANADIEN DES SCIENCES NEUROLOGIQUES
EMG Biofeedback In Patients With Motor Disorders:
An Aid For Co-Ordinating Activity In Antagonistic
Muscle Groups
ALAN E. DAVIS AND ROBERT G. LEE
SUMMARY: A computer program was
developed to analyse the relative amount of
EMG activity in an agonist-antagonist pair
of muscles while subjects performed
voluntary flexion-extension movements at
the wrist to track a visual target. The data
were presented to the subjects in the form
of a vector display, the angle and length of
which was determined from calculation of
EMG power in the two muscles.
This new approach to EMG biofeedback
was evaluated in two hemiplegic patients
and three patients with cerebellar incoordination. Over a training period of several
weeks, all the subjects were able to modify
the pattern of EMG activity in the muscles
to reduce the amount of inappropriate
coactivation of flexors and extensors and
to produce more sustained and regular
activation of individual muscle groups.
RESUME: NOUS avons developpe un
programme d'informatique afin d'analyser
la quantite relative d'activite EMG dans
une paire de muscles agonistes/' antagonistes
alors que les sujets effectuaient des mouvements volontaires de flexion-extension du
poignet defacon a suivre une cible visuelle.
Les donnees furent presentees aux sujets
sous la forme d'un tableau vectoriel dont
I'angle et la longueur etaient determines a
partir du calcul de la puissance EMG dans
les deux muscles.
Cette nouvelle approche de retrocontrole EMG fut evaluee chez deux patients
hemiplegiques et trois patients avec incoordination cerebelleuse. Apres une periode
d'entrainement de plusieurs semaines, tous
les sujets purent modifier le pattern
d'activite EMG dans les muscles etudie's
afin d'ainsi re'duire la quantite de coactivation inappropriee des flechisseurs et
extenseurs et afin de produire une
activation des groupes musculaires individuels soutenue et reguliere.
From The Neuroscience Research Group. Faculty of
Medicine, University of Calgary.
Reprint requests to Dr. Robert G. Lee, Faculty of
Medicine, University of Calgary, Calgary. Alberta,
T2N IN4. Canada.
Vol. 7 No. 3
INTRODUCTION
Biofeedback of physiological parameters of various types has acquired
some popularity in the management of
certain neurologic disorders (see Basmajian, 1979). Several groups of
investigators have employed feedback
of electromyographic (EMG) activity
as an adjunct to conventional physiotherapy in the rehabilitation of motor
disorders such as hemiplegia (Brudny
et al, 1974; Basmajian et al, 1975; Lee
et al, 1976; Basmajian et al, 1977),
torticollis and related dystonic syndromes (Brudny et al, 1976; Korein
and Brudny, 1976; Bird and Cataldo,
1978) and tardive dyskinesia (Albanese and Gaardner, 1977). The usual
approach has been to provide the
patient with a visual or auditory signal
representing E M G activity recorded
from a single muscle. The patient's
task is to either minimize the signal
level to reduce spasticity or involuntary movements or to increase the
signal to reinforce attempts to activate paretic muscles. A limitation of
this "one-dimensional" approach is
that some motor disorders are characterized by defective interaction between opposing groups of muscles (Terzuolo et al, 1973). Under normal
conditions, voluntary activation of a
muscle group is associated with reciprocal inhibition of its antagonists
(Hultborn et al, 1973) but in hemiplegic patients and patients with cerebellar incoordination we have frequently observed examples of abnormal
coactivation of agonist and antagonist muscles. To improve motor
performance these patients are faced
with two tasks. At the same time as
they are activating weak or poorly
coordinated muscles they must learn
to suppress inappropriate activity
from antagonistic muscles.
AUGUST 1980-199
Downloaded from https:/www.cambridge.org/core. IP address: 88.99.165.207, on 14 Jun 2017 at 12:29:40, subject to the Cambridge Core terms of use, available at
https:/www.cambridge.org/core/terms. https://doi.org/10.1017/S0317167100023209
THE CANADIAN JOURNAL OF NEUROLOGICAL SCIENCES
The present study was undertaken
to determine whether biofeedback of a
preprocessed signal providing information concerning the relative amount
of EMG activity in two opposing
muscle groups could assist patients
with motor deficits in carrying out a
simple motor task. A small laboratory
computer system (PDP 11-40) was
used to carry out on-line analysis of
EMG activity from the wrist flexors
and extensors and to generate a visual
display on an oscilloscope. Detailed
studies were carried out on five
patients — two with spastic hemiplegia following cerebral infarcts and
three with cerebellar incoordination.
With feedback training all five patients
were able, over a period of time, to
modify and improve the pattern of
EMG activity in the muscles studied.
MATERIALS AND METHODS
I. Clinical Material
a) Patients with spastic hemiplegia
Case 1: A 39-year-old right handed
woman presented with a subarachnoid hemorrhage due to rupture of a
right internal carotid artery aneurysm. Surgical clipping of the
aneurysm was complicated by postoperative occlusion of the right
internal carotid artery resulting in a
cerebral infarct with severe left
hemiparesis. Some improvement
occurred over the next several
months but at the time biofeedback
studies were commenced 14 months
following the onset of her illness she
still had moderate weakness in most
muscle groups in the left arm,
spasticity in the wrist and finger
flexors, and inability to control
movement of individual fingers.
Attempts to perform rapid flexionextension movements of the wrist
were slow and irregular. Position
sense was diminished in the fingers
of the left hand.
wrist flexion. There was marked
spasticity in the wrist flexors.
Sensation was impaired for all
modalities in the left forearm and
hand,
b) Patients with cerebellar incoordination
Case 3: A 34-year-old left handed
woman presented with a 14 year
history of relapsing and remitting
neurologic symptoms characteristic
of multiple sclerosis. At the time of
our study there was clinical evidence
of lesions at multiple sites including
the optic nerves, brain stem, and
spinal cord. Her major deficit was a
spastic paraplegia, but for several
years she had shown evidence of a
cerebellar deficit in the left arm with
incoordination on the finger-nose
test, mild intention tremor, and
moderate slowing of rapid alternating
movements. Muscle tone in the left
arm was slightly reduced. Sensation
in the upper extremities was normal.
Case 4: A 64-year-old right handed
man had a 15 year history of
progressive spinocerebellar degeneration of undetermined type. He
was confined to a wheelchair because of severe trunkal ataxia.
Examination of the upper extremities revealed severe intention tremor
and incoordination. Rapid alternating movements were extremely slow
and clumsy. Muscle strength in the
arms was normal except for slight
weakness of the intrinsic hand
muscles due to an associated peripheral neuropathy. Sensory examination was normal.
Case 5: A 29-year-old right handed
woman had a 13 year history of
Friedreich's ataxia. She exhibited
marked dysmetria on performing
the finger-nose test with either hand.
Rapid alternating movements were
slow and clumsy. Muscle tone was
reduced and tendon reflexes were
absent in all four extremities.
Position sense was slightly diminished in the fingers.
Case 2: A 51-year-old right handed
man developed a right fronto-parietal
cerebral infarct due to thrombotic
occlusion of the right internal
carotid artery. Initially he had severe
left hemiplegia with complete paralysis of the arm. With intensive
physiotherapy he regained the ability to walk, but there was only
limited recovery in the arm. Biofeedback training was commenced ten
months following the stroke. At that
time the left arm was paralyzed
except for slight movement at the
shoulder and a minimal degree of
2. EMG Recording and Generation of
Display
Recordings were carried out with the
subjects seated in a chair facing the video
terminal (GT 40) of a PDP 11-40
computer. The arm rested on a manipulandum which kept the elbow flexed to
90° and restricted movement to the wrist
joint. The subject's hand was taped firmly
to a handle which permitted voluntary
flexion and extension of the wrist through
a range of approximately 90°. A potentiometer monitored changes in position of
the handle.
200-AUGUST 1980
EMG activity was recorded with pairs of
surface electrodes applied 3.5 cm apart
over the flexor carpi radialis and extensor
carpi ulnaris muscles. After appropriate
amplification, the EMG signals were
sampled on line at a rate of 400 Hz. The
"power" or variance of the signal about a
mean value was computed for each channel
over 100 msec, intervals. The ratio of
flexor EMG power to extensor EMG
power was used to generate a vector display
on the video terminal. Details of the calculations resulting in the vector are provided
in the appendix.
Figure 1 illustrates the display presented
to the subject while carrying out flexion
and extension movements of the left wrist.
When the subject moves the handle to the
right (figure 1-A) flexor EMG power
exceeds extensor EMG power and the
vector is tilted to the right. The opposite
situation is shown in figure 1-C where the
subject is extending the wrist to move the
handle to the left. When the amount of
EMG activity is equal in flexors and
extensors, as in figure I-B, the vector
assumes a vertical direction.
It is important to note than the angle of
the vector is not directly related to handle
position, but rather reflects the relative
degree of activation of flexor and extensor
muscle groups. If the subject activates the
flexors and totally inhibits the extensors
the vector will assume a horizontal position
directed to the right of the screen.
In addition to controlling the direction
of the vector the subject is able to control
its length by varying the total amount of
EMG activity. Thus, a strong contraction
of either or both muscle groups produces a
long vector while complete relaxation
reduces the vector to a very short line or to
a point display.
3. Biofeedback Training Sequence
Training was initiated by having the
computer generate a 4 cm x 4 cm square
target on the video display (figure I, D-F).
The subjects were instructed to experiment
with movements of the handle and
contraction of the forearm muscles until
they were able to locate and hold the end of
the vector within the target area. Once this
was accomplished the computer program
shifted the target 1 mm. to the right and the
sequence was repeated. The target continued to move across the screen in 1 mm.
steps until it reached the edge of the
display, at which time it began to move
back in the opposite direction. Each step in
the sequence occurred only after the
subject had successfully completed the
previous step. The speed at which the target
moved was determined by the accuracy and
skill with which the subject was able to
generate the appropriate level of EMG
EMG Biofeedback and Motor Disorders
Downloaded from https:/www.cambridge.org/core. IP address: 88.99.165.207, on 14 Jun 2017 at 12:29:40, subject to the Cambridge Core terms of use, available at
https:/www.cambridge.org/core/terms. https://doi.org/10.1017/S0317167100023209
LE JOURNAL CANAD1EN DES SCIENCES NEUROLOGIQUES
activity in wrist flexors and extensors to
maintain the vector in the target area.
Simple modifications in the program
permitted the investigator to vary the
complexity of the motor task to match the
needs of individual patients. Although the
velocity of target movement was not
identical for all the patients, it was kept
constant for each individual over the entire
training period.
Following initial familiarization with the
equipment and the video display, patients
returned to the laboratory two or three
times a week for biofeedback training
sessions lasting approximately one hour.
During these sessions the subjects carried
out repeated 5 minute runs of tracking the
visual target, each of which was followed
by a 5 minute session during which they
were instructed to carry out slow wrist
flexion-extension movements at a similar
rate but without visual feedback. Recordings of EMG and handle position during
these "no-feedback" runs were compared
with the data obtained during visual
feedback. Each individual run was followed by a three minute rest period. In
addition, at the conclusion of each session
the subjects were instructed to carry out a
short sequence of flexion-extension movements as rapidly as possible without visual
feedback.
4. Data Analysis
In addition to controlling the EMG
vector display and target movement, the
computer program recorded the time spent
in flexion, extension and coactivation as
the patient tracked the target back and
forth across the screen. Flexion was
defined as the time during which flexor
EMG power was greater than 1.5 times
extensor EMG power and extension was
when extensor EMG power exceeded
flexor EMG power by a similar amount.
Coactivation was defined as the remainder,
that is the time when flexor and extensor
EMG power differed by less than 50%.
iunMwm«ii' —
<*wHWt*
iimU'n • H I » I I
Figure I — Examples of vector display provided to the patients during biofeedback training
of wrist movements and relationship to EMG activity produced by a subject moving his
left wrist. Vector tilts to the right when flexor EMG power exceeds extensor EMG power
(A), to the left when extensor EMG power exceeds flexor EMG power (C). Vector
assumes a vertical position when flexor EMG power equals extensor EMG power. The
length of the vector is determined by the absolute sum of EMG activity in the two
channels.
The computer controlled target which moves back and forth across the display is
illustrated in D-F.
Davis & Lee
RESULTS
Hemiplegic
Patients
Figure 2 shows examples of E M G
activity and associated handle movement for the two hemiplegic patients
during the first recording session and
at the end of a two month training
period. The initial attempts by Case 1
to produce smooth flexion-extension
movements using biofeedback are
characterized by irregular bursts of
E M G activity from the flexors without
any well defined activity from the
extensors (figure 2-A). Similar recording at the end of the training period,
shown on the right, reveals that the
patient has developed the ability to
produce reciprocal activation of flexors and extensors to generate relatively smooth movements of the
handle.
Figure 2-B illustrates attempts by
case 1 to produce rapid flexionextension movements without visual
feedback. During the first session these
movements were slow and irregular
and coactivation of extensors and
flexors occurred commonly. Following the training period the frequency
and rhythmicity of these movements
was increased.
Downloaded from https:/www.cambridge.org/core. IP address: 88.99.165.207, on 14 Jun 2017 at 12:29:40, subject to the Cambridge Core terms of use, available at
https:/www.cambridge.org/core/terms. https://doi.org/10.1017/S0317167100023209
AUGUST 1980-201
THE CANADIAN JOURNAL OF NEUROLOGICAL SCIENCES
FIRST SESSION
EXT EMG
FLEX EMG
LAST SESSION
—H|n ^
I II l>
|
||
|,
POSITION
I
*m
I'll M
M^jkt,
H mi
i \i i mn
— » —
| m
Figure 2 — Recordings of surface EMG from wrist flexors and
extensors and handle position in two hemiplegic patients
performing a wrist flexion-extension task. Recordings on the
left were obtained at thefirstbiofeedback session, those on the
right after two months of feedback training. Tracings A and B
are from case 1. A shows subject's attempt to perform smooth
tracking of the visual target. B represents attempts to carry out
rapid wrist flexion-extension without visual feedback. Tracings
C and D are from case 2 with visual feedback (C) and without
feedback (D). Handle position is not shown for case 2 as he did
not have sufficient strength in the forearm muscles to generate
recordable movement of the handle.
Upward deflection in the handle position trace represents wrist
flexion in this and subsequent illustrations.
Case 2 had marked paresis of the
hemiplegic arm, and although he was
able to activate EMG discharges these
were not sufficient to produce any
appreciable movement of the handle.
Nevertheless EMG recordings with
feedback (figure 2-C) and without
feedback (figure 2-D) show some
improvement in the ability to suppress
coactivation and to selectively control
either flexor or extensor muscles in the
forearm following several weeks of
biofeedback training. The training in
this patient, however, did not result in
significant functional improvement in
control of movement.
To quantitate the effect of biofeedback on the ability of the patients to
control flexor and extensor muscles
independently a ratio was calculated
for each five minute run to indicate the
relative amount of time during which
either flexors or extensors were
predominantly active. The percentage
of total time during which there was
coactivation of flexors and extensors
was also determined. The results for
202-AUGUST 1980
Figure 3 — Relationship between activation of wrist extensors and
wrist flexors and percentage of time during which these muscles
were coactivated for two hemiplegic patients, case 1 (A) and case
2 (B). Each point on the graphs represents results for one five
minute run. Closed circles represent runs with visual feedback,
open circles runs with no feedback. The total time period
represented on the abscissa is approximately three months. The
last five points on each graph were obtained at the one month
followup session after completion of the training sequence.
Value for the ordinates in the upper graphs were obtained by
calculating for each run the following ratio:
Time extensor EMG power is 1.5 times > flexor EMG power
Time flexor EMG power is 1.5 times > extensor EMG power
(see methods)
the two hemiplegic patients over the
two month training period are shown
in figure 3.
Case 1 initially showed a marked
predominance of flexor activation
resulting in very small extensor/flexor
ratios (figure 3-A, top). With biofeedback she was able to produce more
activation of the extensors, and over
the course of the training period the
extensor-flexor ratios gradually approached the optimal value of one
which represents balanced activity in
these opposing muscle groups. This
change was present even in runs
without feedback and was maintained
during the follow-up session one
month later.
The bottom graph in figure 3-A
shows that there was coactivation of
flexors and extensors between 10 and
25% of the total time. This did not
change during the training period
although there was some tendency for
coactivation to be less during runs with
feedback.
In contrast, case 2 showed a gradual
reduction in the proportion of time
during which flexors and extensors
were coactivated, and this effect was
still present at the one month followup
(figure 3-B, bottom). The extensor/
flexor ratios for this patient remained
relatively constant throughout the
recording sessions.
Patients With Cerebellar
Incoordination
Figure 4 shows recordings of EMG
and handle position for case 3, a
patient with multiple sclerosis causing
moderately severe dysmetria of the left
arm. Prior to training (figure 4-A)
bursts of EMG activity were irregular
and prolonged and the resulting
handle movements were slow and
jerky. After eight sessions of feedback
training over a five week period the
patient was able to generate more
discrete bursts of EMG activity and
handle movements were faster and of
greater amplitude. The recordings on
the right of figure 4 show the patient's
EMG Biofeedback and Motor Disorders
Downloaded from https:/www.cambridge.org/core. IP address: 88.99.165.207, on 14 Jun 2017 at 12:29:40, subject to the Cambridge Core terms of use, available at
https:/www.cambridge.org/core/terms. https://doi.org/10.1017/S0317167100023209
LE JOURNAL CANADIEN DES SCIENCES NEUROl.OGIQUES
attempts to carry out rapid wrist
flexion-extension movements without
feedback. The maximum frequency of
these movements was considerably
increased after five weeks of training.
Results for case 4, a patient with
spinocerebellar degeneration and severe intention tremor, are shown in
figure 5. The initial recording (figure 5A) shows inappropriate coactivation
of extensors and flexors with slow
irregular movements of the handle.
After four feedback sessions, each
consisting of three five minute runs,
there was less coactivation and handle
' movements were more regular (Figure
5-B). In addition the patient was able
to maintain this more normal pattern
of movement during voluntary flexion
and extension of the wrist without
feedback (figure 5-C). The effect of
biofeedback on this patient's motor
activity is illustrated further in figure
6-A which illustrates graphically the
relative amounts of flexor and extensor activation and coactivation of both
muscle groups during sequential runs
with and without feedback. The
amount of coactivation is clearly
reduced during runs with feedback.
The data for case 5, a patient with
Friedreich's ataxia, are illustrated in a
similar format in figure 6-B. With this
patient an attempt was made to
determine whether the improved performance with feedback was directly
related to the information provided by
the feedback or whether it might be
due to other associated factors such as
increased attention or motivation. To
investigate this possibility a distorted
form of feedback was introduced on
some runs. The length and angle of the
vector display were randomly scaled
every 100 msec, using a range of plus or
minus 25% of the actual values determined from the calculation of EMG
power. This patient was given the same
instructions as the other patients and
was not informed when the feedback
was being distorted.
Figure 6-B shows that without
feedback this patient had difficulty
producing any activation of flexor
muscles. With distorted feedback there
was some increase in the percentage of
total time during which the flexors
were activated, but this was less than
what occurred when she was provided
with true feedback, which resulted in
the flexors being activated at least 50%
of the time. This suggests that she was
able to use the specific information
provided by the true feedback to assist
her in activating the flexor muscles.
DISCUSSION
These studies show that with visual
feedback of information concerning
relative activation of antagonistic
muscle groups patients with certain
motor deficits are able to modify the
patterns of EMG activity in these
muscles. Patients with either hemiplegia or cerebellar incoordination
were able to utilize the feedback to
reduce the amount of inappropriate
coactivation of flexors and extensors
and to produce more sustained and
regular activation of individual muscle
groups during a simple wrist flexionextension task. The effect of the biofeedback was almost immediate in
some patients, particularly the hemiplegics. They were able to modify their
EMG pattern at the first biofeedback
session, and over a period of time they
were able to maintain this change
independent of feedback. In addition,
the altered pattern of motor activation
was preserved when they were retested
one month following the end of the
twice weekly program of biofeedback
sessions.
It should be emphasized that this
was an uncontrolled study. Although
the interspersion of training runs with
and without visual feedback allowed
each subject to serve in a sense as his
EXT EMG
FLEX EMG
<W<I
ll#
HANDLE
. 20 sec
Figure 4 — EMG and associated handle movements in a patient
with cerebellar incoordination (case 3) before (A) and after (B)
five weeks of biofeedback training. The longer segments of
recording represent slow tracking movements with visual
feedback. The short segments on the right are rapid flexionextension movements performed without feedback.
Davis & Lee
Figure 5 — Case 4 — spinocerebellar degeneration. A) Recording
of EMG and handle position during slow wrist flexionextension movements with visual feedback at the onset of
training. Note the prominent coactivation of flexors and
extensors with attempts to move the handle. B) Recording with
visual feedback after four training sessions. C) Same as B but
without visual feedback.
Downloaded from https:/www.cambridge.org/core. IP address: 88.99.165.207, on 14 Jun 2017 at 12:29:40, subject to the Cambridge Core terms of use, available at
https:/www.cambridge.org/core/terms. https://doi.org/10.1017/S0317167100023209
AUGUST 1980-203
THE CANADIAN JOURNAL OF NEUROLOGICAL SCIENCES
own control, a control group of
subjects who did not receive any
biofeedback training was not studied.
The small number of patients and the
fact that they were not a homogeneous
group with respect to the type of motor
deficit precludes analysis of the data
using conventional statistical methods.
The results indicate that these patients
were able to modify patterns of EMG
activity during biofeedback sessions,
but no conclusions can be reached at
this stage regarding functional improvement or the long-term pratical
advantages of using biofeedback to
treat patients with motor disorders.
Clearly, what is required is a larger
study involving more patients of one
diagnostic category and a control
group which does not receive feedback.
Several questions must be considered before determining whether
these results justify a more extensive
trial of EMG biofeedback of the type
IJ
COACTIVATION
EXTENSION
j.
FEEDBACK
OFF
ON
OFF
ON
j
FLEXION
OFF
B
100
50
o
DISTORTED
Figure 6 — Graphical representation of the percentage of total time allocated to activation
of extensors, activation of flexors, and coactivation (see methods). Each bar represents
EMG activity obtained during a five minute run of wrist flexion-extension movements
with or without visual feedback as indicated at the bottom of the graphs. Results are
shown for two patients with cerebellar incoordination: A - case 4, B - case 5. For case 5
results are also shown for runs using distorted feedback (see text).
we have used. First, was the apparent
improvement in EMG activation directly related to the information
provided by biofeedback or could it
have been due to associated factors?
The exposure to rather complex
electronic and computing equipment
was a novel experience for the patients
and might conceivably introduce a
placebo effect. The attention focused
on the patient and his hemiplegic or
dysmetric arm might also lead to an
increased level of motivation which
could produce the results observed.
Several factors suggest that the
changes observed with biofeedback
were more than a nonspecific placebo
effect. All five patients studied had
stable motor deficits which had not
changed for at least several months
despite prolonged periods of conventional physiotherapy. Also, a clear
difference was noted in the EMG
pattern during sequential runs with
and without feedback. If the changes
were due merely to increased attention
or motivation they should have been
present during the runs without
feedback. Furthermore, case 5 was not
able to perform nearly so well when
provided with distorted feedback,
suggesting that the alteration in EMG
pattern occurring with feedback was
due to the specific information contained in the true feedback signal.
Can EMG biofeedback lead to
improved performance on everyday
motor functions? Obviously, if this
approach is going to be of practical
value in rehabilitation of patients with
motor deficits it must be shown that
the improved patterns of motor
activity which develop during feedback can be transferred to everyday
tasks. This question was not explored
in depth during the present study.
However, four of the five patients did
show definite improvement of the
speed and rhythmicity of rapid voluntary flexion-extension movements of
the wrist carried out without feedback.
Preliminary work has been undertaken
in our laboratory to develop a battery
of functional tests to provide a quantitative assessment of upper extremity
motor control, and it is intended to
incorporate these tests into future
studies to compare motor function
before and after biofeedback training.
Downloaded from https:/www.cambridge.org/core. IP address: 88.99.165.207, on 14 Jun 2017 at 12:29:40, subject to the Cambridge Core terms of use, available at
https:/www.cambridge.org/core/terms. https://doi.org/10.1017/S0317167100023209
204-AUGUST 1980
EMG Biofeedback and Motor Disorders
LE JOURNAL CANADIEN DES SCIENCES NEUROLOGIQUES
Assuming that biofeedback can
result in some modification of patterns
of EMG activation and associated
movements, what are the mechanisms
responsible for these changes? The
simplest explanation is that visual or
auditory feedback provides the central
nervous system with information concerning the moving extremity which is
lacking as a result of the lesion or
disease process. Many normal motor
functions are dependent on continuous feedback of data concerning the
movement which is taking place. The
importance of this feedback becomes
obvious when one observes the striking abnormalities in motor control
which occur in a patient who has lost
proprioceptive sensation as a result of
lesions in the peripheral nerves, dorsal
columns, or somatosensory cortex.
Even with no involvement of efferent
motor pathways, these patients have
considerable difficulty carrying out
coordinated movements, particularly
if they are deprived of visual input. In
this situation feedback of appropriate
information via alternative sensory
pathways should be able to at least
partially substitute for the missing
proprioceptive cues.
But how might biofeedback operate
in other types of motor deficits in
which there are no obvious losses of
proprioception or other sensory modalities? In addition to the direct
feedback pathway discussed above,
there are several other feedback loops
which project to cortical motor
centres. For example the cerebellum,
which receives extensive input from
both the cerebral cortex and from
peripheral receptors, sends projections back to motor cortex, and defects
in this "internal" feedback system are
likely responsible for cerebellar dysmetria (Murphy et al, 1975).
If biofeedback is to be effective one
must determine which particular properties of movement or muscle contraction are most likely to provide
useful information, and how this
information can be processed and
packaged in a format which can be best
utilized by the central nervous system.
In the present study we used a visual
signal which represented the relative
EMG power in agonist and antagonist
muscle groups. Whether this even
remotely resembles the type of infor-
Davis & Lee
mation transmitted to motor control
centers by normal feedback mechanisms is uncertain. Other parameters
which could have been selected for
feedback include position of the limb,
velocity or acceleration of movement,
and force of contraction. Further
studies are required to determine if
some of these parameters might be
more appropriate for feedback than
EMG activity for certain types of
motor disorders.
The possible long-term benefits of
biofeedback therapy have to be
considered in the context of the
mechanisms responsible for spontaneous functional recovery following
an acute focal cerebral lesion such as
an infarct. Early improvement can be
attributed to factors such as resolution of cerebral edema and restoration of function in areas which were
rendered functionally inactive due to
ischemia but which did not undergo
necrosis. However, recovery can continue for many months following a
stroke, and the mechanisms responsible for these late changes are not
well understood. Some of the mechanisms which have been considered
include removal of active inhibitory
influences, collateral sprouting from
surviving axons with formation of new
synapses, transfer of functions to
undamaged areas of cerebral cortex,
the employment of alternative pathways, and the development of new
movement strategies (Luria et al, 1969;
Geschwind, 1974). Whether any form
of biofeedback is capable of influencing
these mechanisms remains entirely
speculative at the moment, but not
outside the realms of possibility.
Two important considerations in
evaluating any new form of therapy
are acceptability to patients and
economic feasibility. In the present
study, all patients adapted readily to
the procedure and to the concept of
interacting with a computer, something
which should not be surprising when
one considers the recent popularity of
electronic games which utilize a television display. The increasing availability of low cost microprocessors and
miniaturized electronic circuits should
make it feasible to carry out more
extensive clinical trials designed to
answer some of the questions which
have been raised concerning the role of
EMG biofeedback in rehabilitation of
motor disorders.
APPENDIX
The computer program for analysis
of EMG activity and generation of the
visual feedback display performed the
following operations. Flexor and
extensor EMG signals were digitized
at a rate of 400 samples per second for
100 millisecond epochs. To measure
power, the variance of each channel
was calculated over the 40 points of the
100 millisecond epoch. For flexor
EMG sample Fi (i= 1, 40) and extensor
EMG sample E/(i= 1,40), the variances
were:
40
VAR(FLEX) = £ (F/-F7) /20
i=l
40
VAR (EXT) = £ (E'-EO /20
i=l
The EMG vector could have been
defined directly in rectangular coordinates using the flexor and extensor
EMG variance as the x and y coordinates. However, this would have
resulted in a vector moving from 0° for
pure flexion to 90° for pure extension.
To make the display more cognitively
useful, the EMG vector was transformed so that 0° represented pure
flexion, while 180° represented pure
extension. This was done by first
converting the EMG vector to polar
coordinates (R, 8):
R = / V A R 2 ( F L E X ) + VAR2(EXT)
8 = TAN"1 [VAR(FLEX)/VAR(EXT)]
The angle 8 was then multiplied by 2 to
give a full display range of 0° to 180°.
The length R and the angle of 28 were
then used to define the EMG vector on
the display. This process was repeated
each 100 milliseconds. Simultaneously
the variances for flexor and extensor
EMG were stored for later analysis, and
the target was stepped across the display
once the EMG vector was positioned by
the subject within the target boundary.
This work was supported by Medical Research
Council of Canada Grant MA 4209. Alan E. Davis was
an M.R.C. Post-Doctoral Fellow during the course of
this investigation.
Downloaded from https:/www.cambridge.org/core. IP address: 88.99.165.207, on 14 Jun 2017 at 12:29:40, subject to the Cambridge Core terms of use, available at
https:/www.cambridge.org/core/terms. https://doi.org/10.1017/S0317167100023209
AUGUST 1980-205
THE C A N A D I A N JOURNAL OF NEUROLOGICAL SCIENCES
REFERENCES
ALBANESE, H. and GAARDNER, K. (1977).
Biofeedback treatment of tardive dyskinesia:
two case reports. Am. J. Psychiat. 134, 11491150.
BASMAJIAN. J.V., KUKULKA, C.G., NARAYAN, M.G. and TAKEBE, K. (1975).
Biofeedback treatment of foot-drop after
stroke compared with standard rehabilitation technique: effects on voluntary control
and strength. Arch. Phys. Med. Rehabil. 56,
231-236.
BASMAJIAN, J. V., REGENOS, E. M. and
BAKER, M. P. (1977). Rehabilitating stroke
patients with biofeedback. Geriatrics 32, 8588.
BASMAJIAN, J.V. (Ed.) (1979). Biofeedback
Principles and Practice for Clinicians.
Williams and Wilkins, Baltimore.
BIRD, B.L. and CATALDO. M. F. (1978).
Experimental analysis of EMG feedback in
treating dystonia. Ann. Neurol. 3, 310-315.
BRUDNY, J., KOREIN, J., LEVIDOW, L, et
al. (1974). Sensory Feedback therapy as a
modality of treatment in central nervous
system disorders of voluntary movement.
Neurology 24, 925-932.
BRUDNY, J., KOREIN, J. and GRYNBAUM,
B.B. (1976). EMG feedback therapy: review
of treatment of 114 patients. Arch. Phys.
Med. Rehabil. 57, 55-61.
GESCHW1ND, N. (1974). Late changes in the
nervous system: an overview. In Plasticity
and Recovery of Function in the Central
Nervous System. Edited by D.G. Stein, J.J.
Rosen and N. Butters, pp. 467-508. Academic Press, New York.
HULTBORN, H., ILLERT, M. and SANTIN1,
M. (1976). Convergence on interneurons
mediating the reciprocal IA inhibition of
motoneurons III. Effects from supraspinal
pathways. Acta Physiol. Scand. 96, 368-391.
KOREIN, J. and BRUDNY, J. (1976).
Integrated EMG feedback in the management of spasmodic torticollis and focal
dystonia: a prospective study of 80 patients.
In The Basal Ganglia. Edited by M. D. Yahr.
pp. 385-424. Raven Press, New York.
LEE, K. H., HILL, E., JOHNSTON, R. and
SMIEHOROWSK1, T. (1976). Myofeedback for muscle retraining in hemiplegic
patients. Arch. Phys. Med. Rehabil. 57, 588591.
LURIA, A. R., NAYDIN, V. L.,TSVETKOVA,
L.S. and VINARSKAYA, E.N. (1969).
Restoration of higher cortical function
following local brain damage. In Handbook
of Clinical Neurology. Edited by P.J.
Winken and G. W. Bruyn. Volume 3. North
Holland, Amsterdam.
MURPHY, J.T., KWAN, H.C., MACKAY,
W.A. and WONG, Y.C. (1975). Physiological basis of cerebellar dysmetria. Canad. J.
Neurol. Sci. 3, 279-284.
TERZUOLO, C.A., SOECHTING, J. F. and
VIV1ANI, P. (1973). Studies on the controls
of some simple motor tasks 11. On the
cerebellar control of movements in relation
to the formulation of intentional commands.
Brain Res. 58, 217-222.
EMG
and
2 0 6 - A Ufrom
G Uhttps:/www.cambridge.org/core.
S T 1980
Downloaded
IP address: 88.99.165.207, on 14 Jun 2017 at 12:29:40, subject to the Cambridge Core
terms Biofeedback
of use, available at
https:/www.cambridge.org/core/terms. https://doi.org/10.1017/S0317167100023209
Motor
Disorders